[R] Did behavior of "paired" argument in t.test() change?

[Robert Baer]

Thanks Ivan and Peter.

I had been extolling the virtues of long format to my students, and 
mentioning some its shortcomings with paired data like paired t.test. I 
then went off script (bad idea) to explain a workaround and fell down 
the rabbit hole. Thebug report 
<https://bugs.r-project.org/show_bug.cgi?id=14359> 14359 on this issue 
was first started in 2010 (several years after I had started using R), 
and finally resulted in a code change only in April 2024 - all of those 
years I was blissfully unaware and had developed some bad habits with 
regard to brittle order assumptions  ;-)

Even after rechecking ?t.test I was not certain that I was remembering 
the old behavior correctly, so it is good to know my memory is intact.  
The help is accurate, but the various pieces are a little dissociated.  
I think I learned I could trust the brittle ordering assumption 
(undocumented) by careful trial and error years ago.

Anyway, thanks for another great R learning experience -- It has been 
fun to see how thoughtfully and carefully the behavior of R base code is 
discussed before a change is made.  I wonder if 14 years of 
thinking/discussion is a record.

Also, I appreciate the suggestion to  use *split()* without having to 
completely ditch the use of the paired = TRUE argument, even if it still 
includes brittle order assumptions for pairs.   I must say that the 
Pair(x,y) ~ 1 formula syntax is not that easy for me to wrap my head 
around. I'll keep watching if there are further changes coming in R-devel

Rob

On 11/5/2025 6:28 AM, Peter Dalgaard wrote:

> Yes. Of course, Pair() requires the data to be in wide format. It's been a while (although not so long ago as some of the other things I cannot remember...) but I think the reason was that the  ...y~g, paired=TRUE... format was considered too brittle with respect to sort order and subsetting.
>
> -pd
>
>
>> On 4 Nov 2025, at 18.40, Ivan Krylov via R-help<r-help using r-project.org> wrote:
>>
>> В Tue, 4 Nov 2025 11:22:57 -0600
>> Robert Baer via R-help<r-help using r-project.org> пишет:
>>
>>> Errorin t.test.formula(value ~ time, data = data_long, paired =
>>> TRUE) :
>>> cannot use 'paired' in formula method
>> This was changed in R-4.4.0:
>> https://bugs.r-project.org/show_bug.cgi?id=14359
>>
>> The news item (see news(grepl('paired', Text))) recommends the use of
>> Pair(x1, x2) ~ 1 instead of the 'paired' argument. A further fix is
>> currently in R-devel (but not R-patched): without it, Pair(x1, x2) does
>> not handle the 'subset' argument.
>>
>> -- 
>> Best regards,
>> Ivan
>>
>> ______________________________________________
>> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guidehttps://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.

-- 
---
Robert W. Baer, Ph.D.
Professor of Physiology
Kirksville College of Osteopathic Medicine
A.T. Still Univerisity of Health Sciences
800 W. Jefferson St.
Kirksville, MO 63501

	[[alternative HTML version deleted]]